Prediction of Drug Positioning for Quan-Du-Zhong Capsules Against Hypertensive Nephropathy Based on the Robustness of Disease Network

Abstract

Hypertensive nephropathy (HN) is a medical condition in which chronic high blood pressure causes different kidney damage, including vascular, glomerular and tubulointerstitial lesions. For HN patients, glomerular and tubulointerstitial lesions occur in different renal structure with distinct mechanisms in the progression of renal damage. As an extraction of Eucommia ulmoides, Quan-du-zhong capsule (QDZJN) has the potential to treat HN due to antihypertensive and renal protective activities. Complicated mechanism of HN underlying various renal lesions and the "multi-component and multi-target" characteristics of QDZJN make identifying drug positioning for various renal lesions of HN complex. Here, we proposed an approach based on drug perturbation of disease network robustness, that is used to assess QDZJN positioning for various HN lesions. Topological characteristics of drug-attacked nodes in disease network were used to evaluated nodes importance to network. To evaluate drug attack on the whole disease network of various HN lesions, the robustness of disease networks before/after drug attack were assessed and compared with null models generated from random networks. We found that potential targets of QDZJN were specifically expressed in the kidneys and tended to participate in the "inflammatory response," "regulation of blood pressure," and "response to LPS and hypoxia," and they were also key factors of HN. Based on network robustness assessment, QDZJN may specifically target glomeruli account to the stronger influence on glomerular network after removal of its potential targets. This prediction strategy of drug positioning is suitable for multi-component drugs based on drug perturbation of disease network robustness for two renal compartments, glomeruli and tubules. A stronger influence on the disease network of glomeruli than of tubules indicated that QDZJN may specifically target glomerular lesion of HN patients and will provide more evidence for precise clinical application of QDZJN against HN. Drug positioning approach we proposed also provides a new strategy for predicting precise clinical use of multi-target drugs.

Keywords: Eucommia ulmoides; drug positioning; hypertensive nephropathy; network pharmacology; network robustness.

FIGURE 1

Tissue enrichment and functional enrichment of potential targets of QDZJN. (A) Histogram of possible target tissue of QDZJN’s potential targets by tissue expression enrichment in DAVID. X-axis is the count of targets, and color gradient of bar is related with p-value of enrichment analysis. (B) Chi-square test table for comparison of the distribution of QDZJN’s targets in disease gene of hypertension and HN. (C) Bubble diagram of enriched GO terms of QDZJN’s targets. X-axis is fold enrichment, Y-axis is related with p-value of enrichment analysis. Bubble size changes with the count of targets in this term.

FIGURE 2

Differentially expressed genes (DEGs) in tubules and glomeruli of HN patients. (A) Volcano plot of differentially expressed genes in tubules and glomeruli. (B) Venn diagram of DEGs in tubules (green) and glomeruli (orange). (C) Bubble plot of enriched GO terms of DEGs of tubules (green) and glomeruli (orange). X-axis is fold enrichment, Y-axis is related with p-value of enrichment analysis. Bubble size changes with the count of targets in this term.

FIGURE 3

Interaction network of DEGs in glomeruli (A) and tubules (B). Node size changes with the node degree in network. The hub nodes were marked with big name labels and potential drug targets of QDZJN were marked with red edges. Nodes with red star label were disease genes related to HN from DisGeNET database.

FIGURE 4

Comparison of topological features of drug-attacked node in tubulointerstitial and glomerular network. (A,B) Boxplot of APL_node (A) and CC_node (B) of drug-attacked nodes and other nodes in tubulointerstitial and glomerular network. ^∗∗p-value < 0.05. (C) Schematic diagram of Permutation test. Network topological features were used as robustness indicators (RI) of networks. Null distribution of RI of 1000 random network was generated to maintain the same edge number of the true network and randomize interactions between nodes. Comparing with the null distribution, statistical significance of RI of true network can be calculated. (D,E) Robustness (APL_net for D, CC_net for E) of glomerular and tubulointerstitial network versus random network (green distribution for tubules and orange distribution for glomeruli). X-axis is the percentage change of robustness indicators before/after attack. ^∗∗p-value < 0.05.

FIGURE 5

Hierarchy network of chemical compound, drug-attacked nodes in the glomerular network and HN related pathological processes. Compounds with red edge represent drug with anti-hypertensive activity.

FIGURE 6

QDZJN against HN via renal protection and reduction of hypertension risk. Left part is the schematic diagram of tissue location of drug based on network robustness. Possible mechanism of QDZJN protect against glomerular injury of HN was summarized in right part.